Skip to main content
. 2012 Sep 24;41(1):e21. doi: 10.1093/nar/gks878

Table 3.

The average performance by different feature selection algorithms on our training data

Feature subsets No. of features Sn (%) Sp (%) Gm (%)
All features (No FS) 125 84.0 98.3 90.86
microPred features (J–M) (10) 21 83.0 97.9 90.14
FS1: ReliefF 50 84.9 98.4 91.40
FS2: InfoGain 75 84.9 98.3 91.35
FS3: CFS + GA 20 84.9 98.6 91.49